Calculation conditions for geometry optimization and energy task.
Since sustainable energy and environment emerging as one of the top issues and challenges for humanity, the photocatalytic hydrogen evolution under visible light has attracted increasing attention. Basically the separation and transmission of photogenerated charge carriers are the two main steps of a photocatalytic reaction. They should be key aspects in the design of efficient photocatalysts for solar energy conversion.
- modified microstructure
- plane‐wave‐density function theory
- hydrogen production
In recent years, energy shortage and environmental pollution have become the focus of world attention. Overuse of fossil fuels is the main reason of these problems. Strong dependence on fossil fuels has also made our economy susceptible to price spikes and intensifying air pollution and global warming. Thus, developing a clean, renewable alternative to fossil fuels is a matter of utmost urgency.
The San Francisco earthquake with magnitude 7.8 in 1906 released an estimated 1017 joules of energy, the amount the Sun delivers to Earth in one second. So earth can get a staggering amount of energy, which was provided by the Sun. This energy can power the great oceanic and atmospheric currents, the cycle of evaporation and condensation of water, and the typhoons, hurricanes, and tornadoes. The 4.6 × 1020 joules, that humans use annually, is equal to the amount of energy that Earth get from the sun in one hour. The 1.2 × 105 terawatts, which Earth gets from the Sun, is far above every other energy source . And solar energy is one of the cleanest energy resources that does not compromise or add to the global warming . It is often called “alternative energy” to fossil fuel energy sources such as oil and coal.
The impressive supply of solar energy is complemented by its versatility, as illustrated in Figure 1. Solar energy applications are divided into three main categories: yield of electricity by exciting electrons in a solar cell, yield of chemical fuel through natural photosynthesis in green plants or artificial photosynthesis in human‐engineered systems, and production of heat using concentrated or unconcentrated sunlight for direct use or further conversion to electricity.
Despite the abundance and versatility of solar energy, we use very little of it to directly power human activities. The reasons are these three main categories have their shortcoming, storage problems for solar electric, emission of greenhouse gases (mainly CO2) for solar fuel and inefficient for solar thermal energies. So we must find ways to store the large quantities of electricity and heat when conversion on a large scale is achieved. Otherwise, solar energy is an interrupted energy, which caused by natural cycles of cloudy‐sunny and day‐night variation, and energy demands are often out of phase with solar energy. Solar fuel production can store energy in chemical bonds. However, it is much more difficult to store electricity and heat. The task, that storing electricity or heat cost effectively for 24 h, is well beyond present technology. So we need a cleaner, more effective, and achieved storage method under present technology.
Hydrogen is potentially an ideal energy carrier, as it is nonpolluting and gives up both its electrons upon oxidation to form only water. It may enable a secure and clean energy future, which uses hydrogen as the energy carrier connecting energy source to diverse end uses. Hydrogen is only an energy carrier for storing and transporting energy, which is different with oil and natural gas. Before we can use hydrogen, we have to produce it since hydrogen does not exist naturally. Every year, more than 500 billion cubic meters are produced [3, 4]. Most of this hydrogen is used for industrial purposes, such as refining petroleum and producing ammonia for fertilizers and other chemicals. In a hydrogen energy system, hydrogen containing compounds such as fossil fuels and water are potential sources of hydrogen [5, 6]. When hydrogen is derived from hydrocarbons, CO2 capture and sequestration are requirements in a low greenhouse gas scenario [7, 8]. However, hydrogen produced from water does not present the challenge of unwanted emissions, but it does require an external resource . If this energy can be obtained from a renewable energy source such as solar energy, hydrogen can then be considered a green energy alternative capable of powering everything. Otherwise, the storage problems of solar energy can be solved by hydrogen. Therefore, achievement of solar hydrogen production from water has been explored vigorously. Several ways for solar hydrogen production are applied (Figure 2) .
Electrolysis of water using a hydroelectric power generation, a solar cell, etc.
Reforming of biomass.
Photoelectrochemical or photocatalytic water splitting.
Since water electrolysis needs a large number of infrastructure constructions for clean power, the practical widespread application of this technique is constrained by its high cost. However, direct solar‐to‐hydrogen conversion based on photochemical and photoelectrochemical water splitting is a promising scheme for sustainable hydrogen production. Currently, direct solar‐to‐hydrogen conversion is in the conceptualization stage still, but ongoing technological progress will ultimately make this approach enter the stage of actual application in the future.
1.1. Fundamental mechanism of photocatalytic hydrogen generation 
Photocatalytic water splitting by solar irradiation is appealing and has been intensively studied in the last four decades [12, 13]. In Fujishima and Honda's pioneering work, they constructed an electrochemical cell for the decomposition of water into hydrogen and oxygen . When the TiO2 electrode was irradiated by light, oxygen evolution occurred at the TiO2 electrode. Concomitant reduction led to hydrogen evolution at the platinum black electrode. After the photocatalyst absorbs photon from light source such as sun, the electrons in the valence band of the photocatalyst are excited to the conduction band, while the holes are left in the valence band (called the negative‐electron (e−) and positive‐hole (h+) pairs). The “band gap” is the energy difference between the valence band and the conduction band. And the light can be effectively absorbed by the photocatalyst, while the wavelength of it corresponds to the “band gap”. After the excited electrons and holes are generated in photocatalyst, they separate and migrate to the surface. Here, they act as reducing agent and oxidizing agent to produce H2 and O2 in the photocatalytic water‐splitting reaction. A schematic representation of the principle of the photocatalytic system for water is depicted in Figure 3.
Water splitting into H2 and O2 is an uphill reaction. It needs the standard Gibbs free energy change ∆G0 of 237 kJ/mol or 1.23 eV, as shown in equation 1.
Therefore, to achieve water splitting, the band gap energy of the photocatalyst should be>1.23 eV (<1000 nm) and it should be <3.0 eV (>400 nm) to use visible light.
The proper band gap and the potentials of the conduction and valence bands are important to facilitate both the reduction and oxidation of H2O. The top level of the valence band has to be more positive than the oxidation potential of O2/H2O (1.23 V), and the bottom level of the conduction band has to be more negative than the reduction potential of H+/H2 (0 V vs NHE).
1.2. The processes of photocatalytic hydrogen evolution
Figure 4 shows the processes in the photocatalytic generation of hydrogen . They include the absorption of light by the photocatalyst, generation of excited charges (electrons and holes), separation of excited charges, recombination of the excited charges, and transfer of excited charges to water or other molecules. The final generation of hydrogen from the semiconductor photocatalyst system can be affected by all of these processes.
The amount of excited electrons in the water/photocatalyst interface mainly determines the total amount of hydrogen generated in reducing water. Obviously, for maximizing the efficiency of the hydrogen generation of the photocatalyst system, any other processes that consume excited electrons should be avoided. Thus, for the charge‐generation process, the reflection or scattering of light by the photocatalyst should be minimized, and the semiconductor photocatalyst should have a low band gap to absorb as much light as possible. Then, the semiconductor photocatalyst should generate excited charges using the absorbed photons, instead of generating phonons or heat.
After excited charges are created in the semiconductor photocatalyst, charge separation/migration and recombination processes are two important competitive processes . The separation of excited electrons and holes may need to overcome an energy barrier. Charge is separated and migrated to the surface of photocatalyst ready for the desired chemical reaction. These processes are beneficial for hydrogen generation through water splitting. However, charge recombination will reduce the efficiency of water splitting since it reduces the amount of charge. It is important for photocatalytic hydrogen generation through water splitting to include efficient charge separation and fast charge transport for avoiding any bulk/surface charge recombination. Any method beneficial to the charge separation and transport should be applied such as design of heterojunction and use of high performance semiconductor materials (photoconductive).
So two keys should be considered to develop a suitable efficiency semiconductor for the visible‐light‐driven photocatalytic splitting of water into H2 and/or O2. Firstly, for harvest visible light and possess the correct band structure, a photocatalyst should have a sufficiently narrow band gap (1.23 eV< Eg < 3.0 eV). Secondly, to avoid bulk/surface electron/hole recombination, photoinduced charges in the photocatalyst should be separated efficiently. In addition, the electron/hole must migrate to the photocatalyst surface for hydrogen and/or oxygen evolution. In summary, the proper photocatalyst with correct band structure and with economical and highly efficient photocatalytic systems for light‐to‐hydrogen energy conversion should be constructed.
1.3. Modifying the electronic band structure for visible‐light harvesting
To develop photocatalyst using visible‐light irradiation, prophase studies were carried out on some narrow band gap semiconductors such as CdS  and WO3[17, 18]. However, the serious photocorrosion of CdS  and the relatively positive conduction band of WO3 proved dissatisfactory for hydrogen production  and created major impediments for visible‐light‐driven water splitting. Some studies were carried out to improve the photocatalytic stability of CdS, [21–24] and others used WO3 as the photoelectrode in the photoelectrochemical cell for water splitting [25–27]. To overcome these disadvantages, many efforts have been made to develop new visible‐light‐driven photocatalysts with high water‐splitting activities [21–27]. Several common approaches have been adopted in order to enhance the performance of visible‐light active of photocatalysts for water splitting, which based on numerous experimental results from the past 20 years. These approaches were shown as follows: (1) developing novel single‐phase Vis‐active photocatalysts through band gap engineering; (2) dye sensitization to make UV‐light‐active photocatalysts harvest visible light; (3) developing solid solutions to control the band structure; and (4) metal or/and nonmetal ions doping for band gap narrowing.
In the method of metal or/and nonmetal ions doping for band gap narrowing, one of the most effective ways to develop visible‐light driven photocatalysts is to create impurity levels in the forbidden band through nonmetal ion doping. It has been widely used to narrow the band gap and improve the visible‐light‐driven photocatalytic activity. Nonmetal‐ion dopants are less likely to form donor levels just like metal‐ion dopants in the forbidden band but instead shift the valence band edge upward. Figure 5 shows a narrowing of band gap. The technology of nonmetal‐ion doping has been widely used to modify some UV‐light‐active oxide photocatalysts, such as Ti‐based oxides [28–30], Ta‐based oxides [31–34], Zr‐based oxides [35, 36], and Nb‐based oxides [37, 38]. The boron doping has been used to modify 3C‐SiC nanowires for effective visible‐light‐driven photocatalytic activity in our recent work (B‐doped 3C‐SiC nanowires) .
Another promising method for controlling photocatalyst band structure is the formation of solid solutions between wide and narrow band gap semiconductors such as doping. According to this method, varying the ratio of the compositions can control the band gap and position. Figure 6 shows the controllable band formation by making a solid solution. This method contain (oxy)sulfide solid solutions [40–42], oxide solid solutions [43, 44], and oxynitride solid solutions [45–48]. The way of oxynitride solid solutions were used in our recent work (TiOxNy/TiN) .
1.4. Efficient photogenerated charge separation
When visible‐light‐driven photocatalysts with proper band structures are synthesized and charges are excited, the photogenerated charge separation is another key factor strongly affecting the efficiency of the photocatalytic water‐splitting process. Clearly, to avoid bulk/surface charge recombination and transfer to the separated active sites on the surface of the photocatalysts, the utilization rate of the photogenerated charges can be improved and high photocatalytic water‐splitting activities can be obtained. On the basis of numerous experimental results and theoretical research results from the past decades, several common approaches have been adopted in order to make efficient photogenerated charge separation: (1) cocatalyst loading; (2) semiconductor combinations; (3) modification of crystal structure and morphology.
Transition metals, especially the noble metals, are widely used as effective cocatalysts for photocatalytic water splitting. Taking Pt as an example, Figure 7 shows the processes of charge transfer between cocatalyst and host photocatalyst. Since the Fermi energy level of noble metal is always lower than that of the semiconductor photocatalyst, the photogenerated electrons migrate to the surface of the host photocatalyst and are entrapped by the noble metal cocatalyst after the surface of the photocatalyst is loaded with the noble metal. Meanwhile, the photogenerated holes migrate to the surface of host photocatalyst. So the photogenerated electrons and holes can be separated efficiently. Subsequently, in the photocatalytic reaction, the separated electrons and holes are involved in the reaction as the reducer and oxidizer, respectively. Overall, the cocatalysts are extremely important. It improves the overall photocatalytic activity of the water splitting because it helps to promote charge separation, which in return reduces both bulk and surface electron/hole recombination. It also accelerates the surface chemical reaction by inhibiting backward reaction. As one of the noble metals, Pt has been widely used as the cocatalyst in photocatalytic water splitting over many different kinds of semiconductors: oxides [50–52], (oxy)sulfides [53–56], and (oxy)nitrides [57–59]. All have been shown to greatly enhance the photocatalytic activity for hydrogen evolution.
The semiconductor combination approach has been shown to be another effective method for improving photocatalytic activity through better photogenerated charge separation with a formation of a heterojunction structure. As shown in Figure 8, the staggered band gap type is the most suitable for photocatalytic applications. In this type, the two semiconductors construct a heterostructure on the basis of the matching band potentials. Holes will transfer to the VB of semiconductor‐A from semiconductor‐B, in which the VB level of semiconductor‐A is higher than semiconductor‐B. Electrons will transfer to the CB of semiconductor‐B from semiconductor‐A, in which the CB level of semiconductor‐A is higher than semiconductor‐B. So the internal field can promote the separation and migration of photogenerated carriers and reduce the recombination of electron–hole. As a result, a mass of holes accumulate on the surface of semiconductor‐A and a mass of electrons accumulate on the surface of semiconductor‐B. These carriers will participate in photoredox reaction and thus the reaction can be enhanced greatly.
In general, the crystal structural features of the materials, such as crystallinity, defects, and any crystal structure distortion, can strongly affect the charge separation and transfer of photogenerated electrons and holes. Additionally, the morphology of the materials is also important. For example, it includes particle size, surface area, surface structure, and active reaction sites. Thus, on the basis of numerous experimental results and theoretical research results from the past decades, both the modification of crystal structure and morphology of the photocatalysts have been adopted in order to improve efficient charge separation. This has had great success in the enhancement of photocatalytic activity for water splitting.
For crystal structure, it has been well demonstrated that the crystal structure of TiO2 plays a significant role in photocatalysis. In photocatalysis, for pure‐phase TiO2, the anatase phase was considered to be more active than the rutile phase [61, 62].
It is well‐known that particle size is a crucial factor in the dynamics of electron/hole recombination (including bulk recombination and surface recombination) processes, especially in semiconductor nanomaterials, and that the movement of electrons and holes is primarily governed by the well‐known quantum confinement . Generally, when the particle size of photocatalysts decreases, its efficiency will increase in photocatalysis [64–67]. Moreover, reduction in particle size could also lead to a larger surface area and increase the available surface active sites [68, 69]. Lee et al. reported that the smaller particle size and higher surface area of NaTaO3 led to higher photocatalytic performance in water splitting . Sathish et al.  found that CdS nanoparticles showed a higher efficiency of photocatalysis compared to bulk CdS. This result indicated the fact that smaller particle size can increase the performance of a photocatalyst. However, when particle size of the nanocrystalline semiconductor becomes extremely small, this in return can offset the benefits of the ultrahigh surface area of the nanocrystalline particles [71, 72]. So for higher efficiency, the photocatalyst should have the proper size.
Two‐dimensional (2D) nanostructures such as nanobelts [73, 74], nanosheets [75, 76], and nanoplates [77, 78] also favor the transfer of electrons and holes generated inside the crystal to the surface and promote charge separation [79–81]. Under identical conditions, Sun et al. found that ZnO nanobelt arrays was better than that found for ZnO film or the rod‐/comblike ZnO nanostructures for the photocatalytic properties . Self‐assembly of nanoscale building blocks into three dimensional (3D) complex structures is another research hot spot in photocatalysis [82–85]. Song and Gao found that hollow NiO microspheres showed a significantly more enhanced photocatalytic activity than NiO rods .
There are two main approaches to enhance the visible-light-driven activity of photocatalysts for developing more efficient that. One is the promotion of the photogenerated charge separation. The other is to narrow the band gaps to harvest visible light in the longer-wavelength regions. Therefore many factors which determine the photocatalytic activity are needed to be applied in one material, such as morphology, structure and crystallinity, chemical composition, surface states, and electronic properties.
Titanium nitride (TiN) is an Fm3m cubic crystal containing both covalent and metallic bonds. It has a narrow energy band gap (0.80 eV) and good electrical conductivity, which are beneficial to photocatalytic H2 production. However, since the rapid recombination between photoelectrons and holes, its application as a photocatalyst is limited. The inherently high defect density of TiN and unfavorable band position for hydrogen reduction led to the rapid recombination of photoelectrons and holes. 3C‐SiC is an important semiconductor with suitable band gap (2.3–3.3 eV) and excellent properties, such as high thermal conductivity, high mechanical strength, and high chemical stability, especially the environmentally friendly property. However, SiC as a photocatalyst does not draw wide attention due to the high recombination rate of photogenerated electron‐hole pairs as well as its surface evolution during 3C‐SiC nanoparticles or nanocrystal contacting with water. In our recent work, three methods including ion doping, combination with other semiconductor to form heterojunction, and 3D complex structures are adopted to improve the photocatalytic activity of 3C‐SiC and TiN from both theoretical and experimental viewpoint, aiming to extend the application of 3C‐SiC and TiN as functional material.
2. Theoretical calculations
Band gap engineering is a significant and new approach, which can modify the band structure of semiconductor. In band gap engineering, solid solution, semiconductor sensitization, and ion doping are the most common ways. What's more, theoretical calculation based on first‐principle calculations can provide an efficient way to identify a candidate and suggest useful processing and production conditions.
Plane‐wave pseudo‐potential with Cambridge Serial Total Energy Package (CASTEP) code is adopted in the first‐principle calculations. The geometry structure is optimized by using the Broyden‐Fletcher‐Goldfarb‐Shanno (BFGS) . Table 1 lists all the calculation conditions. The self‐consistent field is set to 1×10−6 eV/ atom. In the calculation for the DFT exchange‐correlation function, the generalized gradient approximation (GGA) [88, 89]is adopted, as specified by Perdew and Wang . The density mixing option is chosen in the electronic minimization method .
|Task: Geometry optimization||Energy|
|Functional||GGA PW91||LDA CA-PZ [92, 93]|
|Max force:||0.03 eV/||-|
|Max stress:||0.05 GPa||-|
|Max displacement:||0.001 Å||-|
|Algorithm||BFGS:||Use line search||-|
|Stress||0 for all||-|
|Energy cutoff||700 eV||700 eV|
|Pseudopotentials||Ultrasoft ||Norm-conserving |
|FFT grid density||Fine quality||Standard|
|Finite basis correction||Smart||Smart|
|Electronic minimizer||Density mixing||Density mixing|
|k point quality||Fine quality||Fine quality|
|Band structure||Unchecked||Checked||Fine quality k point set|
|Density of states||Unchecked||Checked||Medium quality k point set|
2.1. B‐doped 3C‐SiC nanowires with finned microstructure
3C‐SiC with zinc blende structure is shown in Figure 9(a), and its lattice constant is a=b=c=0.4341 nm. The optimized models are shown in Figure 9(a)–(d), where Si in the symmetrical positions has been substituted by B [96, 97]. The k‐point set mesh parameters are chosen to be fine quality in the Brillouin zone for B‐doped 3C‐SiC (Si31BC32, Si15BC16, and Si7BC8) corresponding to B concentration about 14.3 at%, 6.7 at%, and 3.2 at% respectively.
Figure 10 plots the projected density of states (PDOS) of B‐doped 3C‐SiC with the ratio of B/Si ranging from 0 to 0.143. According to Figure 10a, 3C‐SiC is indirect semi‐conductor, which is in agreement with the reported result . It is changed to direct semiconductor with B being doped (Figure 10(b–d)). As shown in Figure 10e, the band gaps decrease continuously from 2.341 eV to 1.935 eV with the molar ratio of B/Si increasing from 0 to 0.143. It is well known that material with much smaller band gap tends to lead to photo‐corrosion. Dong et al.  reported B‐doped SiC powder for photocatalytic hydrogen evolution under visible light. In their work, when the B/Si molar ratio is 0.05 in B‐doped 3C‐SiC, it exhibits the highest hydrogen evolution rate of 7.41 μmolμg−1 h−1. Therefore in the experiment, the molar ratio of B/Si is selected as 0.067, i.e. Si15BC16.
2.2. TiOxNy/TiN heterojunction composite
As for TiN, TiOxNyis selected as a sensitizer to enhance the photoactivity of TiN. The electronic structures of TiN and TiOxNy can be calculated by plane‐wave‐density function theory (DFT) using the Cambridge Serial Total Energy Package (CASTEP) program package. Figure 11 shows the optimized simple crystal structures of TiN, Ti2O4, and TiOxNy with N doping concentration in the range of 0 to 33 mol% (Ti16O31N, Ti8O15N, Ti4O7N, Ti2O3N, and Ti2O2N2). O in the symmetrical positions of rutile titanium dioxide (TiO2) crystal cell has been substituted by N. TiN has a narrow energy band gap (0.80 eV) in Figure 12a, which is in consistent with Li's et al. work . The calculated band gaps of TiOxNy decrease from 3.084 to 0.009 eV with N doping concentration increasing from 0 to 33 mol% (Figure 12e).
It can be seen that the band gap of TiOxNy is more than 0.8 eV when the doping concentration of N is less than 17 mol% (Figure 12d (Ti2O3N, 0.918 eV)). Since N introduce new energy band, band gaps of TiOxNychanged with N doping concentration. There is a newly formed energy band introduced by the 2p states of doping N in Figure 12(c–d). From above calculation, TiN combining with TiOxNy with proper N doping concentration is expected to form favorable band‐edge position, which may be a potential candidate for visible‐light responsive photocatalyst.
3. Experimental method for water splitting
3.1. B‐doped 3C‐SiC nanowires with finned microstructure
Herein a simple synthesis route, i.e. carbonthermal reduction combining the ion doping and morphology modification is adopted in the present work. By controlling the amount of boron doping and the reaction temperature, B‐doped 3C‐SiC nanowires (NWs) with finned microstructure are obtained. It shows an outstanding activity toward H2 production as high as 108.4 μmol∙h−1, which is about 20 times higher than that of the 3C‐SiC nanowires and 2.6 times of the value reported in the literature .
3.2. TiOxNy/TiN heterojunction composite
TiOxNy/TiN heterojunction composite with tunable chamber structures were prepared through reduction and nitridation of organotitania obtained through solvothermal alcoholysis at 900°C for 4 h in partial cracked NH3. Owing to the low synthesis temperature, TiOxNy/TiN duplicates the original structure of the organotitania. It demonstrates an outstanding photocatalytic activity as high as 34.9 μmol∙h−1∙g−1, which is about 1.5 times higher than the highest value reported in the literature for TiN .
4. Results and discussion
4.1. B‐doped 3C‐SiC nanowires with finned microstructure
4.1.1. Characterization of B‐doped 3C‐SiC NWs
Figure 13 shows the pattern of X‐ray diffraction (XRD) of the B‐doped SiC NWs. For comparison, 3C‐SiC NWs obtained in our experiment is also investigated. In the two sample, three diffraction peaks at 2θ =35.8°, 60°, and 72° corresponding to cubic SiC (JCPDS card no. 73‐1665) both appear suggesting B doping does not change the phase structure of 3C‐SiC. However, the characteristic peak of B‐doped 3C‐SiC shifts towards higher 2θ angle according to the magnification of diffraction peak (111). This change is possibly caused by the substitution of smaller B (0.095 nm) at Si (0.134 nm) . The stacking defects (SF) of SiC were studied by Tateyama et al. . The estimation formula was proposed as follows:
In the formula, I(200) is the intensity values of (200) peak (41.4°) and ISFis the intensity values of SF peak (33.6°). The density of SF in SiC is represented by the ratio value (dSF).As shown in Figure 13, the SF peak (red circles) is too weak to calculate the density, which indicates that there are only a few defects in SiC whiskers.
X‐ray photoelectron spectroscopy (XPS) is used to characterize B‐doped 3C‐SiC NWs. The full spectrum of XPS shows the existence of Si, C, B, F, and O in Figure 14. Since HF is adopted to treat the surface of sample, F possibly comes from it. Figure 14b shows Si 2p fine XPS spectra of the B‐doped 3C‐SiC NWs sample. From the XPS spectra of Si 2p of B‐doped 3C‐SiC (Figure 14b), the peak centered at 100.7 eV corresponds to Si‐C bond in SiC. The peak at 101.8 eV can be assigned to SiOxCy . Two reasons are attributed to the formation of SiOxCy. First is the surface of sample tends to absorb oxygen after treated by HF. Second is SiCxOy is the intermediate product of SiC  that is left on the sample. However, the content of SiOxCy is too low to be detected by XRD analysis. Figure 14c show B 1s fine XPS spectra of the B‐doped 3C‐SiC NWs sample.
XPS signals of B 1s can be assigned to the B‐C bond in SiC, which are observed at 189.8 eV . The peak at 187.4 eV corresponds to the B‐Si bond in SiC and the peak at 193.1 eV corresponds to B‐O bond for B2O3. In this work, these peaks are not found. This indicates that B atoms incorporate into SiC lattice and substitute Si sites, which are consistent with the reported results [96, 97]. The B/Si molar ratio, as determined by XPS, is 0.066, which is close to the theoretical content of Si15BC16. Electron energy loss spectroscopy (EELS) was carried out to further investigate the existence of B element in B‐doped 3C‐SiC. The peak at 284 eV corresponds to C element and the peak centered at 188 eV corresponds to B element (Figure 14d). This further indicates that B‐doped 3C‐SiC is synthesized in the experiment.
SEM and TEM techniques are used to reveal the morphology and microstructure of the as‐prepared samples. Figure 15 show SEM images of the synthesized 3C‐SiC NWs. Figure 15a and d show the overall looks of the 3C‐SiC NWs and B‐doped 3C‐SiC NWs at low magnification, which indicate that the majority of the nanowires can be described as long and straight filaments with the length up to several mm. As shown in SEM at higher magnification (Figure 15b), the typical morphology of 3C‐SiC NWs possess smooth surface and the average diameter is about 80 nm. The whisker has a homogeneous crystalline structure with fringes spacing at 2.51 Å as shown in Figure 16b, which is characteristic of 3C‐SiC. By comparison, the morphology of B‐doped 3C‐SiC NWs changes a lot. The typical morphology of the B‐doped 3C‐SiC NWs is finned nanowires as shown in the magnified images (Figure 15e and f). It is composed of inner core stems and outer fins. Figure 16c and d show TEM images of typical finned like B‐doped 3C‐SiC NWs. The thickness of the fins is 10–20 nm. The diameter of the inner core stems is about 80 nm and the fins is about 100–200 nm. The lattice spacing of a fin is 2.50 Å as shown in Figure 16e. From SAED pattern of the fin (the inset of Figure 16e), it is single crystal. The HRTEM image reveals that the lattice spacing of the nanowire is also 2.50 Å in Figure 16f. It is caused by substitution of smaller B (0.095 nm) at Si (0.134 nm) and leads to the distortion of the lattice.
For the formation of hierarchical nanostructures, two growth mechanisms are put forward [103–105]. Self‐assembly of nanobuilding blocks is one formation mechanism, wherein blocks include balls, wires, and platelets . Inner 1D core structures and then the epitaxial growth of secondary branches is another formation mechanism [103, 104]. As shown in Figure 17a, some SiC NWs possessing rough and uneven surface are found in this work. Therefore the formation of finned 3C‐SiC NWs cloud be the two‐step epitaxial growth process.
In the experiment, the possible reactions take place as following:
SiO vapor plays an important role during the synthesis of 3C‐SiC whiskers . The formation of B according to Equation (2) makes B doping possible and thus is a key step. The standard reaction Gibbs energies of Equation (2) are 767412.5 J∙mol−1using the database of FactSage 6.4. In our experiment, the partial pressure of O2 is controlled to be 2.6×10−12 Pa, which is verified by mass spectrometry. The vapor pressure of B2O3 is 152 Pa at 1500°C according to the Speiser's  result as following:
Therefore, the reaction Gibbs energy of Equation (2) can be calculated as follows:
where J is the reaction constant of Equation (2) at 1500°C. Therefore the reaction Gibbs energy of Equation (2) is calculated to be −58715.5 J/mol, indicating that the reaction can take place under the experimental condition.
Owing to the high temperature, the formation of inner 1D core stem is the first step in a two‐step epitaxial growth process. When reaction between CO, B, and SiO occurs, the B‐doped SiC(Si15BC16) will be formed (Equation (3)). CO2 reacts with C to form CO according to Equation (4), which causes CO to remain at a supersaturated level and thus Equation (3) continues to proceed. This leads B‐doped SiC(Si15BC16) to precipitate as nuclei and grow along the (111) direction due to the lowest surface energy to form the core stem B‐doped SiC (Si15BC16) nanowires. Since SiC NWs have more defects after B doping, the nanowires possess rough and uneven surface (as shown in Figure 17a). When the temperature is decreased and the amount of SiO becomes less, the growth rate of B‐doped SiC will become lower. Since the energy for nucleating SiC on B‐doped SiC core stems is far lower than that on the other places, the newly formed B‐doped SiC tends to grow on that. Therefore, the fins are formed on the core stems. With time prolonging, B‐doped SiC nanowire with finned structure are produced.
4.1.2. Water splitting for hydrogen
The 3C‐SiC NWs and B‐doped 3C‐SiC NWs are evaluated under visible‐light irradiation for their photocatalytic activities for H2‐production in aqueous suspensions with Na2S and Na2SO3 as sacrificial agents (electron donor). Photocatalytic activities of B‐doped SiC with different B/Si ratios are compared in Figure 18. B doping significantly affects the photocatalytic activity as shown in Figure 18. The photocatalytic H2‐production rate of sample from high to low is Si15BC16 (108.4 μmol∙h−1∙g−1), Si7BC8 (81.0 μmol∙h−1∙g−1), Si31BC32 (62.1 μmol∙h−1∙g−1), and pure 3C‐SiC (5.46 μmol∙h−1∙g−1).The recycling reaction experiments were carried out to investigate the stability of photocatalytic hydrogen production of 3C‐SiC NWs and B‐doped 3C‐SiC NWs (Si15BC16) under visible light irradiation. No decrease in catalytic activity is observed in the recycling reactions as shown in Figure 18b. The H2 production rate for the B‐doped 3C‐SiC NWs exceeds that of 3C‐SiC NWs by more than 20 times. It also exhibits enhanced activity toward H2 production compared with the recent work as shown in Figure 18c. Wang's work group reported SiC fine powder, which can achieve at a value of 1.11 μmol∙h−1∙g−1. Guo's work group reported modified SiC nanowires, which can achieve at a value of 2.68 μmol∙h−1∙g−1. And Dong's work group reported boron‐doped SiC powder, which can achieve at a value of 7.41 μmol∙h−1∙g−1. Among these work, the reduced grapheme oxide/SiC showed the larger visible‐light‐driven photocatalytic hydrogen production done by Yuan's work group to be 42.4 μmol∙h−1∙g−1. Our result is 108.4 μmol∙h−1∙g−1, which is 2.6 times higher than that of the reported reduced grapheme oxide/SiC.
In view of the photocatalytic hydrogen production mechanism of B‐doped 3C‐SiC NWs, first is attributed to the smaller band gaps with B doping as shown in Figure 10. It is well know that the positions of the valence band maximum (VBM) and the conduction band minimum (CBM) are critical variables in determining the feasibility of visible‐light‐driven photocatalytic hydrogen production. As shown in Figure 19, it can be seen that the CB and VB potentials (ECB and EVB) of 3C‐SiC are ECB=0 eV and EVB= 2.341 eV (Figure 19a and b). After B doping, the CB and VB potentials (ECB and EVB) are decreased to ECB=−0.4 eV and EVB= 1.827 eV (Figure 19c and d). Since the mixing of Si 3p and B 2p orbital, the CBM of B‐doping 3C‐SiC decrease. And since the mixing of C 2s, Si 3s, and B 2p orbital, the VBM decrease. As a result, the CB edge potential of B‐doped 3C‐SiC is more negative than normal hydrogen electrode (NHE), indicating that photoinduced electrons can easily transfer from B‐doped 3C‐SiC NWs to H+ and generate hydrogen. This can also be confirmed by the UV‐vis diffusion reflectance spectra (Figure 20a). The UV‐vis spectra of 3C‐SiC NWs shows a strong ultraviolet absorption from 200 to 500 nm (the black line of Figure 20a). The red line of Figure 20a is the UV‐vis spectra of B‐doped 3C‐SiC NWs, which the absorption edge extends to the visible region (about 800 nm) and the absorption intensity is also enhanced. The band gap of the SiC is calculated using the Kubelka‐Munk function . As shown in Figure 20b, an intercept value of 2.34 eV for 3C‐SiC NWs (the black line) is close to that of the bulk 3C‐SiC (2.35 eV). An intercept value of 2.216 eV for B‐doped 3C‐SiC NWs (the red line) is close to that of the calculated value using DFT.
During the photocatalytic process, the photo generated electron recombination is mainly responsible for the low photocatalysis efficiency. Charge transfer effect can be characterized by PL quenching effect within the sample. After B doping, the PL intensity of 3C‐SiC decreases obviously (Figure 21). PL is mainly caused by the transitions of electrons to release photons. In this work, electron is transferred instead of transitioning to lower energy level in the B doping 3C‐SiC NWs. Otherwise, as shown in Figure 13a, less stacking defects and the special finned like single crystalline nanowires are beneficial to enhance transfer of electron. This makes the charge pathway much smoother. Since the multiple light reflects within the fins of nanowires (inset Figure 22) and increases the contacting area with aqueous suspensions, the special finned like morphology also contributes to higher photocatalytic hydrogen production.
From above results, using a facile and simple carbothermal reduction of a mixture of low‐cost boric oxide powder, gangue and carbon black, finned nanostructure B‐doped 3C‐SiC NWs were synthesized. The growth process belongs to a two‐step epitaxial growth mechanism. Combining the theoretical calculation and experimental result, B substitutes Si during the doping process and possesses smaller band gaps. B‐doped 3C‐SiC NWs demonstrates an enhanced and stable activity for H2 production as high as 108.4 μmol∙h−1, which is about 20 times of that of 3C‐SiC and 2.6 times of the value reported in the literature. It appears that the combined effect of such factors as the single crystal, electronic structures, and the finned like morphology is attributed to the enhanced photocatalytic hydrogen production. This work presents an applicable method to develop various semiconductors with controllable morphology applied in solar energy conversion, gas sensors, and photoluminescence.
4.2. TiOxNy/TiN heterojunction composite
4.2.1. Crystal phase analyses
In all the samples, five diffraction peaks at 2θ = 36.90°, 42.86°, 62.24°, 74.61°, and 78.48° corresponding to TiN (PDF 38‐1420).As shown in Figure 23a, with increasing nitridation temperature, the crystallization degree and purity of TiN increase. The constant a increases from 4.216 (800°C) to 4.241 Å (900°C) according to the data of refinement XRD (Table 2). And the theoretical value of TiO is 4.180 Å and that of TiO is 4.180 Å . The value of sample is between TiO and TiN, indicating the existence of TiOxNy. When the temperature reaches up to 950–1000°C, the constant a is calculated to be 4.241Å, which is in consistent with the standard value of TiN (PDF 38‐1420 (a=4.241Å)) .
|Temperature (°C)||800||850||900||950||1000||JCPDS card 38-1420|
|Grain size (nm)||35.5±1.5||36.1±1.9||37.0±1.6||37.9±1.7||38.5±2.4|
In view of the effect of reaction time, Figure 23b shows the XRD patterns of the samples obtained at 900°C for different reaction time at partial cracked NH3 atmosphere. When the reaction time is 3 h, some small diffraction peaks of TiO2 still exist. All the characteristic peaks are indexed to be that of TiN with the nitridation time is extended to 4–5 h. Table 3 shows the refinement of XRD data at 900°C for different time. The constant a increases with the reaction time prolonging. When the nitridation time is 4 h, the constant a (4.238 Å) is between the theoretical value of TiO (4.180 Å) and TiN (4.241 Å), indicating the existence of TiOxNy. From above experiments, the synthesis condition of TiOxNy/TiN composite is selected as 900°C and 4 h. Pure TiN can be produced at 1000°C for 3 h.
|Time (h)||2||3||4||5||JCPDS card 38-1420|
|Grain size (nm)||35.2±1.3||36.8±2.2||37.2±1.4||38.3±0.7|
XPS is adopted to further characterize TiOxNy/TiN composite. Figure 24(a) shows the full XPS spectrum of the samples obtained at 900°C for 4 h, which indicate the existence of C, N, Ti, and O. Figure 24(b–d) show fine XPS spectra of Ti 2p, N 1s, and O 1s. The Ti 2p bands have three groups of peaks, i.e. the typical peaks for TiN (455.3 eV, 457.5 eV, 460.9 eV, and 463.0 eV), the peaks for TiOxNy(456.1 eV and 461.7 eV), and the peaks for TiO2 (459.1 eV and 464.7 eV) (Figure 24b) [110–113]. The existence of TiO2is attributed to the reaction of TiN and TiOxNy particles with oxygen from air atmosphere. From the XPS spectra of N 1s (Figure 24c), the peak at 396.3 eV can be assigned to TiOxNy [110, 111, 114]. The peak centered at 397.3 eV corresponds to Ti‐N bond in TiN and peak at 399.2 eV corresponds to C=N bond . The absorption of C from the atmosphere led to the existence of C=N. From the XPS spectra of N 1s, the molar content of N in TiOxNy is about 6.8 mol%. XPS signals of O 1s (Figure 24d) are observed at around 530.4 eV and 532.0 eV, which can be assigned to Ti‐O bond in TiO2 and TiOxNy respectively [110, 111], indicating that the samples are composed of TiN and TiOxNy. The molar ratio of TiOxNy/TiN is determined by XPS to be 15.64, corresponding to the mass content of TiN to be 95 mass%. Since the content of TiOxNy was too low to be detected by XRD, these are not the diffraction peaks of TiOxNy XRD patterns.
4.2.2. Microstructure analyses
SEM images of TiN/TiOxNy obtained at 900°C for 4 h in partial cracked NH3 are shown in Figure 25. It can be seen that the microstructure of TiOxNy/TiN nanoparticles duplicate the original structure of the precursors. The average diameter of TiOxNy/TiN composite with solid microstructure is about 1–2 μm (Figure 25b). As shown in Figure 25d, yolk‐shell TiOxNy/TiN are composed of a thin shell with an average thickness around 20 nm and a small yolk with an average diameter around 500 nm. As shown in Figure 25f, a thin shell with an average diameter around 1–2 μm and an average thickness around 20 nm combine the hollow microspheres.
TEM images of TiOxNy/TiN composite are shown in Figure 26. The low‐magnification TEM images (Figure 26a and b) show that TiOxNy/TiN composite is composed of regular aggregated particles. As shown in Figure 26c, two particles firmly stick together, in which one lattice spacing is 0.40 nm and the other is 0.25 nm. The lattice spacing of two particles is consistent with the (111) planes of TiOxNy and the (111) planes of cubic TiN.
In addition, it can be seen that TiN crystal is in close contact with TiOxNy crystal to form a heterojunction, which is believed to promote the transfer of photogenerated electrons and holes between TiN and TiOxNy, suppresses their recombination, and thus enhances the photocatalytic activity [115–117].
4.2.3. Photocatalytic activities
The water splitting of H2O to H2 performance of solid TiOxNy/TiN, yolk‐shell TiOxNy/TiN and hollow TiOxNy/TiN samples under light irradiation was examined in the presence of Na2S and Na2SO3 as an hole acceptor. For comparison, the photocatalytic H2 production of TiN and TiOxNy(N/Ti=2) is also carried out respectively. TiN was synthesized at 1000°C for 3 h in flowing partial cracked NH3. TiOxNy (N/Ti=2) were synthesized through a microemulsion‐hydrothermal method using triethylamine, urea, thiourea, and hydrazine as organic compounds . The mixture is put in Teflon‐lined stainless steel autoclave and heated at 120°C for 12 h. The results are shown in Figure 27. By comparison, the as‐prepared TiOxNy/TiN heterojunction with yolk‐shell and hollow structures exhibit better photocatalytic H2 production activities, while the maximum H2 production rate is achieved for yolk‐shell TiOxNy/TiN with 35 μmol∙h−1∙g−1. The yolk‐shell TiN synthesized at 800°C for 20 h was reported by Li's et al. group, which showed the larger visible‐light‐driven photocatalytic hydrogen production to be 22.8 μmol∙h−1∙g−1. Our result is 1.5 times of that value. The recycling reaction experiments were carried out to investigate the stability of photocatalytic hydrogen production of yolk‐shell TiOxNy/TiN under visible light irradiation. No decrease in catalytic activity is observed in the recycling reactions as shown in Figure 27b.
The enhanced photocatalytic hydrogen production performance of the yolk‐shell TiOxNy/TiN heterojunction is attributed to the formation of heterojunction by TiN and TiOxNy with the proper electronic structures and its special nanostructures. From the theoretical calculation as shown in Figure 12, TiN combining with TiOxNy with proper N doping concentration is expected to form favorable band‐edge position, which may be a potential candidate for visible‐light responsive photocatalyst. From XPS spectra, the molar content of N in TiOxNy is about 6.8 mol%, corresponding to the band gap of 1.56 eV. When TiN and TiOxNy is combined together, the band gap should be in the range of 0.8 and 1.56 eV. UV‐Vis diffuse reflection spectra (Figure 28) is also carried out. The absorption band‐edge for yolk‐shell TiOxNy/TiN are at about 850 nm, corresponding to band gap energy to be 1.4 eV, verifying the theoretical calculation.
Under visible‐light irradiation, the photogenerated electrons are excited from the valence bands (VBs) to the conduction bands (CBs) of TiN and TiOxNy in the TiOxNy/TiN composite, creating positive holes in VB of TiN and TiOxNy (Figure 29a). Since CB (−0.6 eV)12 level of TiN is lower than that of TiOxNy (CB: −0.4 eV) , electrons in CB of TiN can be transferred to that of TiOxNy. The VB (0.2 eV)  level of TiN is lower than that of TiOxNy (VB 0.6–1.1 eV) , holes in VB of TiOxNy can be transferred to that of TiN. Therefore, the probability of electron–hole recombination can be reduced, which can be confirmed by the photoluminescence spectra (Figure 29b). It is found that TiOxNy/TiN exhibits a fluorescence decrease (or quenching) as compared with TiN and TiOxNy , indicating that the photogenerated carrier recombination is inhibited greatly. These results should be derived from the imitate contacts between TiN and TiOxNy. Finally, the sacrificial reagents oxidized by the positive holes in the surface of TiN and the photo‐excited electrons in the surface of TiOxNy can generate hydrogen, and thus the photocatalytic reaction can be enhanced greatly .
Another important factor for the enhancement of the photocatalytic activities is the special yolk‐shell nanostructures. First, the specific surface area of yolk‐shell nanostructure is increased significantly, which increase the aqueous suspension contact area. Second, since the resulting yolk‐shell nanostructures, there are the multiple light reflections within the chamber, which leads to efficient photocatalytic and photoelectro‐chemical performances.
Therefore from above results, TiOxNy/TiN heterojunction composite with solid, yolk‐shell, and hollow microspheres were prepared through reduction and nitridation of organotitania obtained through solvothermal alcoholysis. Especially, the yolk‐shell TiOxNy/TiN heterojunction demonstrated an outstanding activity toward H2 production as high as 34.9 μmol∙h−1∙g−1, which was about 1.5 times of the value reported in the literature. The electronic structure of heterojunction and the yolk‐shell morphology are beneficial to enhance the H2 production activity.
5. Summary and prospects
In our work including B‐doped 3C‐SiC nanowires and TiOxNy/TiN heterojunction composite, the electronic band structure, the crystallinity and crystal structure, surface structure, and morphology of photocatalysts on a nanometric scale had been investigated systematically and modified to optimize the preparation methods and refine the materials for maximizing efficiency. B‐doped 3C‐SiC NWs demonstrates an enhanced and stable activity for H2 production as high as 108.4 μmol∙h−1, which is about 20 times of that of 3C‐SiC, and 2.6 times of the value reported in the literature . And the yolk‐shell TiOxNy/TiN heterojunction demonstrated an outstanding activity toward H2 production as high as 34.9 μmol∙h−1∙g−1, which was about 1.5 times of the value reported in the literature . However, these improvements are not enough for industrial applications. Since the low photocatalytic efficiency and the lack of extensive studies for a successful scale‐up of the laboratory setup into an industrially relevant scale. Therefore in the viewpoint of material and reactor design, reduction of cost will have to be given special priority before the final utilization of semiconductor‐based photocatalytic hydrogen generation .
This study was supported by the National Science Fund for Excellent Young Scholars of China (No. 51522402), the National Natural Science Foundation of China (Grant No. 51572019)and the Central Universities of No. FRF‐TP‐13‐006A and FRF‐TP‐15‐006C1.